Technology still manages to shock me. Not because it’s clever. We all know it’s clever. What surprises me is how often it fails at basic things.
I’m not talking about complicated scientific systems or experimental gadgets. I mean the everyday stuff. The things we all rely on without thinking.
And oddly enough, the more advanced everything becomes, the less dependable it sometimes feels.
I’ve written before about older technology being simple and reliable. Not perfect. Often crude. But it worked. Radios played. Telephones rang. Televisions showed the picture without needing five updates and a firmware patch.
Today we’ve got devices that would’ve looked like science fiction thirty years ago. Yet we still sit there watching a spinning buffering circle like a bunch of patient fools.
When Sound and Picture Can’t Agree With Each Other
Take something as basic as watching a video.
In our house alone there are multiple ways to do that. A seriously powerful PC, an M4 Mac. A MacBook Pro. An M3 iPad. A few iPhones. Each one capable of connecting to wired headphones, Bluetooth headphones, wired speakers, Bluetooth speakers. Bluetooth itself now has endless versions – 5.0, 5.1, and so on.
Wi-Fi has evolved as well. The newer iPad I replaced my old one with, for example, has noticeably stronger Wi-Fi capability than the 2020 model.
And yet.
Buffering still happens.
Audio drifts out of sync with video.
Bluetooth adds that tiny delay (latency) which becomes painfully obvious once you notice it.
My son and I were watching football recently and the moment a player struck the ball, you could see the kick happen before you heard the sound – or hear it before seeing it (I don’t remember which way round it was). A fraction of a second perhaps. But once you see it, you can’t unsee it.
So you exit the stream. Go back in again. Sometimes it fixes it. Sometimes it doesn’t. I bet we’ve all done this dance more times than we can count.
Even the idea of watching something live has become questionable. You think you’re watching a live broadcast. Then someone in the next room cheers ten seconds before the goal appears on your screen.
Not catastrophic problems. Just constant little irritations that shouldn’t exist in systems this advanced.
The Odd Fragility of Modern Infrastructure
The modern digital world runs on layers of invisible dependencies.
- Your device talks to your router.
- Your router talks to your internet provider.
- That provider talks to servers scattered around the planet.
- The video stream travels through content delivery networks.
- The audio gets encoded, decoded, compressed, decompressed.
All of that happens in seconds.
Which is impressive. But it also means there are dozens of points where things can wobble.
A weak Wi-Fi signal. Congestion on a network. Bluetooth latency. A slow server somewhere you’ve never heard of.
And suddenly your perfectly modern setup behaves like a dodgy VHS player from the 1980s.
AI Has the Same Reliability Problem
Artificial intelligence has its own version of this.
We’ve now had more than three years of mainstream AI systems since tools like ChatGPT appeared in late 2022. During that time we’ve heard endless discussions about hallucinations.
That polite word for when the system confidently invents something that isn’t true.
It doesn’t happen as often as it used to. The systems have improved. They check more sources. They cross-reference information. They are far better at grounding answers in evidence.
But there’s still a fundamental issue.
They refuse to say “I don’t know.”
Instead they try to answer anyway.
Sometimes the answer is correct. Sometimes it’s close. And occasionally it’s completely wrong but delivered with perfect confidence.
Anyone who’s used these systems regularly has seen it happen.
You ask a question. The system responds as if it’s absolutely certain. Then you check the details and realise parts of it simply don’t exist.
Which brings us right back to reliability.
An encyclopaedia from twenty years ago might have been limited. Out of date even. But if something was printed on the page, you could usually trust that someone had checked it.
Modern systems can generate answers instantly. But instant answers aren’t always accurate ones.
The Strange Contrast With The Past
This is where the whole thing becomes slightly surreal.
In 1969, during the Apollo 11 Moon landing, humans managed to hold a conversation with astronauts roughly 240,000 miles away on the Moon.
Yet today we can struggle to stream a football match without the audio (or video) arriving late.
Of course the comparison isn’t entirely fair. Different systems. Different engineering challenges.
Still, it highlights something uncomfortable. Technological progress doesn’t automatically mean reliability improves at the same pace.
- Sometimes complexity introduces new weaknesses.
- Sometimes convenience replaces robustness.
- Sometimes the system becomes so intricate that small problems ripple through the whole thing.
And that leaves us in a strange position.
We live in the most technologically capable era in history. Devices in our pockets have more computing power than entire buildings once did.
Yet every so often
- the Wi-Fi drops out,
- the stream buffers,
- the picture lags behind the sound, or
- the clever AI assistant calmly tells us something that simply isn’t true.
Which makes you wonder whether the real challenge for modern technology isn’t building smarter systems.
It’s building ones we can actually trust.
Questions Worth Asking About Modern Technology
Why does Bluetooth audio sometimes go out of sync with video?
Bluetooth introduces a small delay while the audio signal is compressed, transmitted, and decoded by the receiving device. Even a delay of a few milliseconds can cause noticeable lip-sync issues, particularly with video or live sports.
Why do live streams often run behind real time?
Streaming platforms buffer video to prevent interruptions when network speeds vary. That buffering creates a delay, often somewhere between 5 and 30 seconds, so what you’re watching is already slightly in the past.
What does it mean when AI “hallucinates” an answer?
It’s when a language model generates something that sounds completely plausible but is wrong, or simply made up. The system predicts likely words and patterns rather than verifying facts, which is why confident answers aren’t always correct ones.
Has AI reliability improved since 2022?
Yes, meaningfully. Large language models have got significantly better at reducing hallucinations and cross-referencing sources. That said, they can still produce confident but incorrect responses, so verifying anything important independently remains good practice.
Why does modern technology sometimes feel less reliable than older devices?
Older technology was simpler, with fewer things that could go wrong. Modern systems depend on multiple interconnected layers – software, cloud services, wireless networks, and hardware all talking to each other. More capability, but also more places for something to quietly break.